|
| 1 | +import os.path |
| 2 | +from typing import Optional |
| 3 | + |
| 4 | +import gradio as gr |
| 5 | + |
| 6 | +import lcc.cleaner |
| 7 | + |
| 8 | +# --------------------------------------------------------------------------- |
| 9 | + |
| 10 | + |
| 11 | +READONLY_PATHS = os.getenv("READONLY_PATHS", "True").lower() == "true" |
| 12 | + |
| 13 | + |
| 14 | +# --------------------------------------------------------------------------- |
| 15 | + |
| 16 | + |
| 17 | +with gr.Blocks() as cleaner: |
| 18 | + dn_resources = "resources/cleaner" |
| 19 | + |
| 20 | + # configuration |
| 21 | + with gr.Accordion("⚙️ Sentence Cleaner Configuration", open=False): |
| 22 | + dn_rules = gr.Textbox( |
| 23 | + value=dn_resources, |
| 24 | + label="Folder with cleaner rule files", |
| 25 | + max_lines=1, |
| 26 | + interactive=not READONLY_PATHS, |
| 27 | + visible=not READONLY_PATHS, |
| 28 | + ) |
| 29 | + fn_replacements = gr.Textbox( |
| 30 | + value="StringReplacements.list", |
| 31 | + label="String replacement mapping file (filename in rules folder)", |
| 32 | + max_lines=1, |
| 33 | + interactive=not READONLY_PATHS, |
| 34 | + visible=not READONLY_PATHS, |
| 35 | + ) |
| 36 | + |
| 37 | + with gr.Row(): |
| 38 | + text_type = gr.Textbox( |
| 39 | + label="Text type, e.g., 'web', 'news'/'newscrawl', 'wikipedia', ...", |
| 40 | + max_lines=1, |
| 41 | + interactive=True, |
| 42 | + ) |
| 43 | + lang_code = gr.Textbox( |
| 44 | + label="Language of text, e.g., 'deu', 'eng', ...", |
| 45 | + max_lines=1, |
| 46 | + interactive=True, |
| 47 | + ) |
| 48 | + |
| 49 | + do_replacements = gr.Checkbox( |
| 50 | + value=True, |
| 51 | + label="Whether to apply text replacements before checking sentence quality", |
| 52 | + interactive=True, |
| 53 | + ) |
| 54 | + |
| 55 | + # inputs |
| 56 | + document_text = gr.Textbox( |
| 57 | + lines=3, label="Text", placeholder="Enter a single sentence" |
| 58 | + ) |
| 59 | + |
| 60 | + # action buttons |
| 61 | + cleaner_btn = gr.Button("Sanitize Sentences", variant="primary") |
| 62 | + |
| 63 | + # outputs |
| 64 | + with gr.Group(): |
| 65 | + status_text = gr.HTML(label="Sentence Quality Status") |
| 66 | + replaced_text = gr.Textbox( |
| 67 | + label="Text with replacements", show_copy_button=True |
| 68 | + ) |
| 69 | + filter_details_json = gr.JSON(label="Filter rules that flagged this sentence") |
| 70 | + |
| 71 | + # worker function |
| 72 | + def clean_text( |
| 73 | + text: str, |
| 74 | + dn_rules: Optional[str], |
| 75 | + text_type: Optional[str] = None, |
| 76 | + lang_code: Optional[str] = None, |
| 77 | + fn_replacements: Optional[str] = None, |
| 78 | + do_replacements: Optional[bool] = True, |
| 79 | + ): |
| 80 | + cwd = os.getcwd() |
| 81 | + if os.path.relpath(os.path.realpath(dn_rules), cwd).startswith("../"): |
| 82 | + raise gr.Error("Invalid 'dn_rules' path!") |
| 83 | + if fn_replacements and os.path.relpath( |
| 84 | + os.path.realpath(os.path.join(dn_rules, fn_replacements)), cwd |
| 85 | + ).startswith("../"): |
| 86 | + raise gr.Error("Invalid 'fn_replacements' filename!") |
| 87 | + |
| 88 | + if not os.path.isdir(dn_rules): |
| 89 | + gr.Warning("Folder 'dn_rules' not found!") |
| 90 | + if not os.path.isfile(os.path.join(dn_rules, fn_replacements)): |
| 91 | + gr.Warning("File 'fn_replacements' not found!") |
| 92 | + |
| 93 | + lcc_cleaner = lcc.cleaner.SentenceCleaner( |
| 94 | + dn_rules=dn_rules, |
| 95 | + text_type=text_type, |
| 96 | + lang_code=lang_code, |
| 97 | + fn_replacements=fn_replacements, |
| 98 | + ) |
| 99 | + |
| 100 | + replaced = lcc_cleaner.replacer.replace(text) if do_replacements else None |
| 101 | + results = lcc_cleaner.filter_sentence_results( |
| 102 | + text, do_replacements=do_replacements |
| 103 | + ) |
| 104 | + filter_details = [ |
| 105 | + {"rule": filter.id_, "description": filter.description, "hit": hit} |
| 106 | + for filter, hit in results.items() |
| 107 | + if hit |
| 108 | + ] |
| 109 | + |
| 110 | + is_ok = not filter_details |
| 111 | + |
| 112 | + status = "✅ Sentence is ok." if is_ok else "❎ Sentence is bad!" |
| 113 | + new_status_text = gr.HTML( |
| 114 | + value=f"<p style='margin: 1rem; text-align: center; font-weight: bold;'>{status}</p>" |
| 115 | + ) |
| 116 | + |
| 117 | + return { |
| 118 | + replaced_text: replaced, |
| 119 | + filter_details_json: filter_details if filter_details else None, |
| 120 | + status_text: new_status_text, |
| 121 | + } |
| 122 | + |
| 123 | + # action buttons event handler |
| 124 | + cleaner_btn.click( |
| 125 | + fn=clean_text, |
| 126 | + inputs=[ |
| 127 | + document_text, |
| 128 | + dn_rules, |
| 129 | + text_type, |
| 130 | + lang_code, |
| 131 | + fn_replacements, |
| 132 | + do_replacements, |
| 133 | + ], |
| 134 | + outputs=[replaced_text, filter_details_json, status_text], |
| 135 | + ) |
| 136 | + |
| 137 | + |
| 138 | +# --------------------------------------------------------------------------- |
| 139 | + |
| 140 | +if __name__ == "__main__": |
| 141 | + cleaner.launch(show_api=False, share=False) |
0 commit comments